In today's information age, the competitive edge of many companies and public trust in government institutions can depend on the security of the information held in its databases. Breaches of that security are a highly topical issue for both designers and users of database systems.
Therefore, it is desirable to provide a secure method to both identify and authenticate users of information services. Conventional biometric systems provide a partial solution to the aforementioned need. As used herein, the term "biometric" refers to the automated process of determining positive identification on the information contained within one or more of an individual's unique physiological characteristics. In general, a biometric authentication system includes a statistical model for a particular user, such as a mixture Gaussian speech model. A person is considered to be authenticated if the system provides a score below a rejection threshold. Other persons in the population, due to physiological differences, score much more poorly and likely appear outside the acceptance region. The biometric statistical model is represented using a set of statistical parameters, such as the average spectrum. The statistical parameters are typically stored in a data structure as part of a database used by the authentication algorithm. Fingerprints, hand geometry, voice pattern, retinal pattern, iris scans, signatures and others all constitute sources of unique physiological characteristic which can establish identity.
Traditional biometric systems store their biometric information in databases in an unencrypted form. There are drawbacks in storing information in such manner. Whenever information is stored in unencrypted form this situation can lead to any number of planned attacks by prospective unauthorized individuals. It may be possible, for example, for an unauthorized user to copy a parameter of an authorized user in such a way as to gain access to a system. For example, if the database was stolen by a prospective attacker, the attacker would be able to choose the most closely matching statistical model in the database and claim to be that person.
Therefore, there is a need for a system that stores biometric information in a secure manner so as to prevent the occurrence of theft and attacks from unauthorized personnel.